Hi, I'm Federico Garzón. I connect a company's memory, an agent that talks with every leader (your company's Jarvis) and a game that moves the team every day on WhatsApp. Financial, operational and commercial decisions under one roof, with AI and strategy.
Most companies buy AI tools one by one. I build the whole loop: what the company knows feeds the agent, the agent talks with leaders, the game moves the team and every decision goes back into memory.
Meetings, decisions, processes and data in one brain that people and agents read.
Stage-weighted close forecast, cash, investor data room and a debt-funding map.
Where each loan is stuck today, which queue to hit first and what to automate next.
Which channel brings real business, which lead is unanswered and who to call now.
Every number carries a label. Measured: I can point to the log, the counter or the dashboard. Company result: it happened while the system was live, but I don't claim the system caused it.
It's the company's Jarvis: it watches the loan pipeline, drafts follow-ups and answers ops questions on WhatsApp, with per-user permissions and human approval before every send.
Ianna pre-qualifies a mortgage against 7 banks' policies, reads borrower documents and takes calls. Prompt caching cut the cost per conversation 10x.
Fixed indexing on the main site and built a tracker of how ChatGPT, Claude, Gemini and Perplexity mention the brand.
Shared inbox, unanswered-lead alerts, a voice agent that answers and transfers, Meta ad attribution and card payments. Three brands use it today.
LQN World gives 30+ people across 8 areas their queue, their goal for the day and their team's podium, with WhatsApp summaries twice a day and an automatic month-end close.
I co-founded and ran DORA: underwriting with KYC, insurance and vintage-based portfolio risk. 2,000 tenants in 15 cities, profitable and acquired. It's why I build AI the way I do: rules in code, every decision reproducible.
For fintechs, banks, insurers, real estate firms, developers and proptechs whose operations live in WhatsApp, documents and credit.
I map your processes, find the three AI use cases with the best return and leave you a 90-day plan with costs, risks and metrics.
A real agent working with your team: on WhatsApp, by voice or reviewing documents. With tests, permissions and human approval from day one.
I sit with your leadership team on a recurring basis. AI governance, roadmap, team hiring and a dashboard showing what AI does to the P&L.
For boards and leadership teams: what's real, what's hype and how to start without burning cash.
Plus: access to my network of investors, debt funds, AI talent and LatAm clients, when it serves your plan.
Book 30 minutesThe job is to take most of them back. Rules that matter live in code, below the model, where a prompt can't negotiate them.
An agent earns the right to act alone with evidence, not enthusiasm.
Anonymized contracts, never other people's inboxes.
Proposes who to contact; never guesses an address.
The agent writes; a person sends.
With brakes and the owner's sign-off: if nobody switches it on, nothing goes out.
Only with evidence: a long record approved unchanged and zero incidents.
If a rule matters, enforce it below the model.
Verify the side effect, not the return code.
Monitoring must not depend on the thing it monitors.
The model explains; the code decides.
Fifteen real incidents, each with the rule it left behind. It's what nobody shows in a demo, and what helps most when you're about to put AI to work. Headlines here; the detail, in conversation.
A prompt is a request; the transport layer is a wall.
A dashboard showing zero should be an error, not a number.
Verify the side effect, not the return code.
Anything that can create a commercial promise is computed, not generated.
Voice agents fail in the room, not in the lab.
Before a metric moves money, measure how reliable it is.
Co-living: 15.7K-member community, 105 tenants, 4.8/5.
SaaS for real estate agents: MRR +12% a month.
OKRs and operating rhythm for startups growing ~10% a month.
Rental guarantees. 2,000 tenants, 0.8% default. Acquired.
InmoLawyer in production and a go-to-market agent in 3 days for a DTC brand.
Production agents, a voice CRM and gamified ops for a VC-backed mortgage platform, plus a plugin published in the ChatGPT directory.
AI is going to be a commodity.La República · 2026
Book 30 minutes: tell me the process that hurts most and I'll tell you, no strings attached, where AI would move your numbers first. Or email me; I reply personally within 48 hours.
